On Intervals, Transitivity = Choas

Author(s): Michel Vellekoop and Raoul Berglund

Source: The American Mathematical Monthly , Apr., 1994, Vol. 101, No. 4 (Apr., 1994),
pp. 353-355
Published by: Taylor & Francis, Ltd. on behalf of the Mathematical Association of
America

Stable URL: https://www.jstor.org/stable/2975629

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Taylor & Francis, Ltd. and Mathematical Association of America are collaborating with JSTOR
to digitize, preserve and extend access to The American Mathematical Monthly

This content downloaded from

189.68.23.153 on Thu, 25 Apr 2024 21:53:39 +00:00
All use subject to https://about.jstor.org/terms
 This content downloaded from
189.68.23.153 on Thu, 25 Apr 2024 21:53:39 +00:00
All use subject to https://about.jstor.org/terms
positive value in z and a negative value in g(z) and this would mean, by the
Intermediate Value Theorem, that there exists a point c c ]z, g(z)[C J with
gk(C) - C = 0, giving a km-periodic point of f in J. Thus z < gk(Z) for all
positive k so in particular for k = n - m > 0, giving z < f(n-m)m(z). Since we
assumed that f n m(f m(z)) < f m(z) we could prove analogously, taking g = f nrn
that f (n-m)mr(f m(z)) < f m(z). But then we have that the function f (n-m)m(x) -x
has a positive value in z and a negative value in f m(z), giving an (n - m)m-periodic
point in J and thus a contradiction. The other case can be proven analogously. El

Proof of the Proposition: Suppose that f is continuous and topologically transitive.

Because of the result in [2] we only need to prove that the periodic points are
dense in I. Suppose that this is not the case, then there exists an interval J c I
containing no periodic points. Take an x c J which is not an endpoint of J, an
open neighbourhood N c J of x and an open interval E c J \ N. Since f is
topologically transitive on I there exists a natural number m > 0 with f m(N) n E
A 0 and thus a y c J with fm(y) c E c J. Since J contains no periodic points we
know that y # fm(y) and since f is continuous this implies that we can find a
neighbourhood U of y with ftm(U) n U = 0. Since U is an open set we can use
the topological transitivity again and find an n > m and a z c U with f n(z) c U.
But then we have 0 < m < n and z, f n(z) E U while f m(z) 4 U and this violates
our earlier lemma. El
We know now that for maps on an interval the only condition that has to be
checked for Devaney's definition of chaos is the first one, topological transitivity.
Note that the proof cannot be generalized for higher dimensions or the unit circle
S1 because our lemma uses the ordering on R in an essential way.
For completeness we note that there are no other trivialities in Devaney's
definition when restricted to intervals:

A continuous function on an interval whose periodic points are dense doesn't need
to have sensitive dependence on initial conditions.

The identity function on any interval trivially proves this.

A continuous function on an interval which has sensitive dependence on initial

conditions and whose periodic points are dense does not have to be transitive.

Define on I = R+the function

{3x 0 <x < 3

f (x) 3x - 2 +22<4 X
-3x <X <<12
1

(f(x - )+ I x2 1
It is sensitive on initial conditions since Idf/dx(x)l = 3 for all points on I, so e
neighbourhood around a point will expand under iteration. It is easy to establish
that f n has 3n - 2 fixed points between any two integer values with distances
between these points smaller than (y)n-1, so the periodic points are dense. But
since f([O, 1]) = [0, lIthe function is not topologically transitive. When one re-
stricts this function to the interval I = [0, 2] one sees that it is a counterexample
for finite I as well.

354 NOTES [April

This content downloaded from

189.68.23.153 on Thu, 25 Apr 2024 21:53:39 +00:00
All use subject to https://about.jstor.org/terms
 A continuous function on an interval which has sensitive dependence on initial
conditions doesn't need to have periodic points which are dense.

As counterexample take the interval I = [[O, ] and the function

f3x 0 ?X<
f(x) 3 { (1 -X) I <X < 3
Sensitive dependence is clear again since the func
be no periodic points in 10, 3/8[ since it is easy to e
initial value in this subinterval, will not return t
infinite case, take I= R +and f(x) = 2x.

REFERENCES

1. R. Devaney, An Introduction to Chaotic Dynamical Systems, Addison-Wesley, 1989.

2. J. Banks, J. Brooks, G. Cairns, G. Davis and P. Stacey, On Devaney's definition of Chaos,
American Mathematical Monthly, 99 (1992) 332-334.
3. D. Assaf, IV and S. Gadbois, Definition of chaos, letter in American Mathematical Monthly, 99
(1992) 865.
4. L. S. Block and W. A. Coppel, Dynamics in One Dimension, Lecture Notes in Mathematics no.
1513, Springer-Verlag 1992.

Dept. of Applied Mathematics Department of Mathematics

University of Twente Abo Akademi
Enschede, The Netherlands Abo, Finland

Proof of a Mixed Arithmetic-Mean,

Geometric-Mean Inequality

Kiran Kedlaya*

The following conjecture was made by F. Holland in [3].

Conjecture. Let xl, x2,..., x,, be positive real numbers. The arithmetic mean of the
numbers
3 n

x1x x, xlX2X3,. ., VXlX2 Xn

does not exceed the geometric mean of the numbers

x1 + X2 X1 + X2 + X3 X1 + X2 + ... +Xn
1 2 ' 3 '' n

There is equality if a

*The author wishes to tha

1994] NOTES 355

This content downloaded from

189.68.23.153 on Thu, 25 Apr 2024 21:53:39 +00:00
All use subject to https://about.jstor.org/terms